Dimensionality reduction and visualization techniques t-stochastic neighbour embedding (t-SNE) and uniform manifold approximation and projection (UMAP) were used to evaluate National Renewable Energy Laboratory’s (NREL) market segmentation for rooftop solar technical potential based on small, medium, and large classification labels. The medium and large class clusters were shown to overlap over a broad range of hyperparameter optimizations leading to the agglomeration of both classes and a revised dataset with binary classifications, small and large. T-SNE outputs in the low-dimensional embedded feature space were used as inputs for support vector machine classification algorithms. While the polynomial kernel trick performed poorly at classifying the distinct binary clusters, both the radial basis function and sigmoid kernel tricks precisely and accurately classified the new rooftop solar technical potential market segments.